Re: st: Marginsplot on backtransformed data

[Ulrich Kohler <ukohler@uni-potsdam.de>]

Here is a solution that works reasonably well (using a log transform).


. use http://www.stata-press.com/data/kk3/data1, clear
. gen logincome = log(income)
. reg lsat logincome i.sex i.mar
. margins, at(logincome=(6(.1)10))
. matrix at = r(at)
. parmest, bmat(r(b)) vmat(r(V)) fast
. svmat at
. gen x = exp(at1)
. tw || rarea min95 max95 x || line estimate x

data1.dta is the prime example data set of Kohler/Kreuter "Data Analysis 
Using Stata", published by Stata Press, and -parmest- is uswer-written 
by Roger Newson and available from SSC:

. ssc install parmest

Many regards

Uli


Am 19.12.2013 07:58, schrieb Daniel Herbert Opi:
> Hi All,
>
> I am using the marginsplot function to plot graphs that take into
> account adjustments I have carried out following a regression
> analysis. Since my regression analysis contains data that has been
> normalised by square root transformation, the plots are not that
> easily interpretable. Is there a way to plot the same graphs but
> somehow tell Stata/marginsplot to use back-transformed results? And
> does the margins function store predictions that can be accessed?
>
> I am using Stata 13 and the following are examples of commands I am using:
>
> regress squareroot_transformed_variable i.categorical1 i.categorical2
>
> margins categorical1
>
> marginsplot
>
> Kind Regards
> Herbert
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/